Tell Perl that your data files are in iso-8859-7 when you read them in (i.e. open them with "<:encoding(iso-8859-7)"), and tell Perl that your input received from the browser via CGI is in utf-8 (my $word = Encode::decode_utf8(param('word'));, or similar — in case it's not already flagged as utf8 (which is hard to tell without seeing the code) ). This results in both sides of what you want to match to be decoded into Perl character strings, which will handle wide characters correctly.

(If you can't get it to work, please show the actual code you've tried; that makes it easier to help...)